COVID-19 Outbreak: Worldwide analysis

Quick overview

Current status

#library(nCov2019)
library(leaflet)
library(dplyr)
library(ggplot2)
library(plotly)
library(scales)
library(xts)
library(dygraphs)
library(corrplot)

COVID<-read.csv("covid_19_data.csv")
COVID_2<-read.csv("COVID19_7-Apr.csv")

Format date:

Date<-as.Date(COVID_2$Date, format="%m/%d/%y") 

COVID_2$Date2<-Date

COVID_updated<-COVID_2 %>% filter(Date2==max(Date2))

leaflet(width = "100%") %>% 
  addProviderTiles("CartoDB.DarkMatter") %>% 
  setView(lng = 0, lat = 10, zoom = 1.5) %>% 
  addCircleMarkers(data = COVID_updated, 
                   lng = ~ Long,
                   lat = ~ Lat,
                   radius = ~ log(Confirmed+1),
                   color = rgb(218/255,65/255,56/255),
                   fillOpacity = ~ ifelse(Confirmed > 0, 1, 0),
                   stroke = FALSE,
                   label = ~ paste(Province.State,",",Country.Region, ": ", Confirmed)
                   )

Current top 10 countries:

COVID_top<-COVID_2 %>% filter(Date2==max(Date2)) %>% 
  group_by(Country.Region) %>% summarise(Total_confirmed=sum(Confirmed)) %>% 
  top_n(10,Total_confirmed) %>% arrange(desc(Total_confirmed))

plot<-ggplot(data=COVID_top
       , aes(x=Total_confirmed,y=reorder(Country.Region,Total_confirmed))) +
  geom_bar(stat ="identity",alpha=0.8,fill="firebrick3") +
  geom_text(aes(label=Total_confirmed), vjust=0.5, hjust=0.9,color="black", size=3.5) +
  scale_x_continuous(labels = comma) +
  labs(title = paste("Top 10 countries with confirmed cases as of ",max(COVID_2$Date2)),
       x = "Confirmed cases",
       y = "Country") +
  theme_minimal()

ggplotly(plot,tooltip = c("x"),width=750)

Time distribution:

COVID_2_Day<- COVID_2 %>% group_by(Date2) %>% summarise(World_confirmed=sum(Confirmed),
                                                        World_deaths=sum(Deaths),
                                                        World_recovered=sum(Recovered))


COVID_Day_confirmed_series<-xts(COVID_2_Day$World_confirmed, order.by=COVID_2_Day$Date2)
COVID_Day_deaths_series<-xts(COVID_2_Day$World_deaths, order.by=COVID_2_Day$Date2)
COVID_Day_recovered_series<-xts(COVID_2_Day$World_recovered, order.by=COVID_2_Day$Date2)

Day_summary<-cbind(COVID_Day_confirmed_series,COVID_Day_deaths_series,COVID_Day_recovered_series)

dygraph(Day_summary, main = "SARS-COV2-outbreak: Total worldwide cases", 
        xlab="Date", ylab="Total cases",width = 750) %>% 
  dySeries("COVID_Day_confirmed_series", "Total cases",drawPoints = TRUE, 
           pointSize = 3, color=rgb(53/255,116/255,199/255)) %>% 
  dySeries("COVID_Day_deaths_series", "Total deaths",drawPoints = TRUE, 
           pointSize = 3, color=rgb(189/255,55/255,48/255)) %>% 
  dySeries("COVID_Day_recovered_series", "Total recovered",drawPoints = TRUE, 
           pointSize = 3, color=rgb(69/255,136/255,51/255)) %>% 
  dyRangeSelector()

Team members countries total cases:

COVID_2_Day_Lebanon<- COVID_2 %>% 
  filter(Country.Region %in% c("Lebanon")) %>% 
  group_by(Date2) %>% summarise(World_confirmed=sum(Confirmed))

COVID_2_Day_Chile<- COVID_2 %>% 
  filter(Country.Region %in% c("Chile")) %>% 
  group_by(Date2) %>% summarise(World_confirmed=sum(Confirmed))

COVID_2_Day_Colombia<- COVID_2 %>% 
  filter(Country.Region %in% c("Colombia")) %>% 
  group_by(Date2) %>% summarise(World_confirmed=sum(Confirmed))

COVID_2_Day_CostaRica<- COVID_2 %>% 
  filter(Country.Region %in% c("Costa Rica")) %>% 
  group_by(Date2) %>% summarise(World_confirmed=sum(Confirmed))


COVID_Day_series_Lebanon<-xts(COVID_2_Day_Lebanon$World_confirmed, order.by=COVID_2_Day_Lebanon$Date2)
COVID_Day_series_Chile<-xts(COVID_2_Day_Chile$World_confirmed, order.by=COVID_2_Day_Chile$Date2)
COVID_Day_series_Colombia<-xts(COVID_2_Day_Colombia$World_confirmed, order.by=COVID_2_Day_Colombia$Date2)
COVID_Day_series_CostaRica<-xts(COVID_2_Day_CostaRica$World_confirmed, order.by=COVID_2_Day_CostaRica$Date2)

Our_Countries<-cbind(COVID_Day_series_Lebanon,COVID_Day_series_Chile,COVID_Day_series_Colombia,COVID_Day_series_CostaRica)

dygraph(Our_Countries, main = "SARS-COV2-outbreak: Total cases by country", xlab="Date", ylab="Total cases",width = 750) %>% 
  dySeries("COVID_Day_series_Lebanon", "Lebanon",drawPoints = TRUE, 
           pointSize = 3, color=rgb(0,0,3/255)) %>% 
  dySeries("COVID_Day_series_Chile", "Chile",drawPoints = TRUE, 
           pointSize = 3,color=rgb(120/255,28/255,109/255)) %>% 
  dySeries("COVID_Day_series_Colombia", "Colombia",drawPoints = TRUE, 
           pointSize = 3,color=rgb(237/255,105/255,37/255)) %>% 
  dySeries("COVID_Day_series_CostaRica", "Costa Rica",drawPoints = TRUE,
           pointSize = 3,color=rgb(204/255,197/255,126/255)) %>% 
  dyRangeSelector()

Looking for correlations

fig <- plot_ly(COVID_updated, x = ~Confirmed, y = ~Deaths, z = ~Recovered, width=750) %>% 
  add_markers(text= ~Country.Region ,hoverinfo= "text",
              marker = list(color=rgb(189/255,55/255,48/255))) %>% 
  layout(title="Confirmed cases Vs. Deaths Vs. Recovered", scene = list(
                    xaxis = list(title = 'Confirmed'),
                     yaxis = list(title = 'Deaths'),
                     zaxis = list(title = 'Recovered'))) 
fig

Human Development Index

HDI<-read.csv("Human Development Index (HDI)_2.csv",sep=";",dec=",")

COVID_Country<-COVID_2 %>% filter(Date2==max(Date2)) %>% 
  group_by(Country.Region) %>% summarise(Total_confirmed=sum(Confirmed),
                                         Total_deaths=sum(Deaths),
                                         Total_Recovered=sum(Recovered))

Remove after parentheses:

HDI$Country_2<-gsub("\\s*\\([^\\)]+\\)","",as.character(HDI$Country))

HDI$Country_2[HDI$Country_2=="United States"]<-"US"
HDI$Country_2[HDI$Country_2=="Korea"]<-"South Korea"

Population:

Population<-read.csv("World_population.csv",sep=";",dec=",")

Remove after commma:

Population$Country_Name_2<-gsub(",.*", "", as.character(Population$Country_Name))

Population$Country_Name_2[Population$Country_Name_2=="United States"]<-"US"
Population$Country_Name_2[Population$Country_Code=="KOR"]<-"South Korea"
Population$Country_Name_2[Population$Country_Code=="CZE"]<-"Czechia"

Natural Join:

COVID_3<- COVID_Country %>% inner_join(HDI,by=c("Country.Region"="Country_2")) %>% 
  inner_join(Population,by=c("Country.Region"="Country_Name_2")) %>% 
  select(Country.Region,Total_confirmed,Total_deaths,Total_Recovered,HDI_Rank_2018,Year_2018,
         Country_Code,Population_2018) %>% 
  mutate(Cases_million=(Total_confirmed/Population_2018)*1000000,
         Recovered_percentage=(Total_Recovered/Total_confirmed)*100)  

COVID_3<-COVID_3[!is.na(COVID_3$Population_2018),]

Plot the Human Development Index(HDI) Vs. the number of cases (applying a log transformation), and the proportion of recovered cases:

plot<-ggplot(data=COVID_3,aes(x=log(Cases_million),y=Year_2018,
                        size=Recovered_percentage,text=Country.Region)) +
  geom_point(color="black",fill=rgb(237/255,105/255,37/255),shape=21,alpha=0.6) +
  scale_size(range = c(3,15), name="Recovered \n percentage") +
  theme_minimal() + 
  theme(legend.position="bottom") +
  labs(title="HDI Vs. logarithmus of COVID-19 cases by million inhabitants \n and proportion of recovered",
       x="ln(Cases/1M population)",
       y="HDI")

ggplotly(plot,tooltip = c("text"),width=750)

COVID_numeric_1<-COVID_3 %>% mutate(Log_cases=log(Cases_million),
                                    Death_percentage=(Total_deaths/Total_confirmed)*100) %>% 
  select(Log_cases,Recovered_percentage,Death_percentage,Year_2018)

corrplot(cor(COVID_numeric_1),method = "number",tl.col="black",tl.srt=15,
         col=colorRampPalette(c(rgb(204/255,197/255,126/255),rgb(237/255,105/255,37/255)))(200))

Infants lacking immunization, measles (% of one-year-olds)

Measles<-read.csv("Measles_immunization.csv",sep=";",dec=".")

Measles$Country_2<-gsub("\\s*\\([^\\)]+\\)","",as.character(Measles$Country))

Measles$Country_2[Measles$Country_2=="United States"]<-"US"
Measles$Country_2[Measles$Country_2=="Korea"]<-"South Korea"

COVID_3<- COVID_3 %>% inner_join(Measles,by=c("Country.Region"="Country_2")) %>% select(-c("Country"))

#write.csv(COVID_3,"COVID_Covariables.csv")

ggplot(COVID_3, aes(y=Measles_2018)) + 
  geom_boxplot(fill="dodgerblue4",outlier.shape = 21, 
               outlier.fill = "firebrick",alpha=0.75) +
  ggtitle("Boxplot of % infants lacking measles immunization") + ylab("% of infants") +
  theme_minimal()

ggplot(COVID_3, aes(Measles_2018)) + 
  geom_histogram(fill="dodgerblue4",bins=20,alpha=0.8) +
  ggtitle("Histogram of % infants lacking measles immunization") + 
  xlab("% of infants") + 
  ylab("Count") +
  theme_minimal()

plot<-ggplot(data=COVID_3,aes(x=log(Cases_million),y=Measles_2018,
                        size=Recovered_percentage,text=Country.Region)) +
  geom_point(color="black",fill=rgb(120/255,28/255,109/255),shape=21,alpha=0.6) +
  scale_size(range = c(3,15), name="Recovered \n percentage") +
  theme_minimal() + 
  theme(legend.position="bottom") +
  labs(title="% infants lacking measles immunization Vs. logarithmus of COVID-19 cases by million inhabitants \n and proportion of recovered",
       x="ln(Cases/1M population)",
       y="% of infants")

ggplotly(plot,tooltip = c("text"),width=750)

Health expenditure (% of GDP)

Health_expenditure<-read.csv("Health_expenditure_GDP.csv",sep=";",dec=".")

Health_expenditure$Country_2<-gsub("\\s*\\([^\\)]+\\)","",
                                   as.character(Health_expenditure$Country))

Health_expenditure$Country_2[Health_expenditure$Country_2=="United States"]<-"US"
Health_expenditure$Country_2[Health_expenditure$Country_2=="Korea"]<-"South Korea"

COVID_3<- COVID_3 %>% inner_join(Health_expenditure,by=c("Country.Region"="Country_2")) %>% select(-c("Country"))

#write.csv(COVID_3,"COVID_Covariables.csv")

plot<-ggplot(data=COVID_3,aes(x=log(Cases_million),y=Expenditure_2016,
                        size=Recovered_percentage,text=Country.Region)) +
  geom_point(color="black",fill=rgb(204/255,197/255,126/255),shape=21,alpha=0.6) +
  scale_size(range = c(3,15), name="Recovered \n percentage") +
  theme_minimal() + 
  theme(legend.position="bottom") +
  labs(title="Health expenditure (% of GDP) Vs. logarithmus of COVID-19 cases by million inhabitants \n and proportion of recovered",
       x="ln(Cases/1M population)",
       y="% of GDP in health")

ggplotly(plot,tooltip = c("text"),width=750)

Fitting a regression model

Mod1<-lm(log(COVID_3$Cases_million)~COVID_3$Year_2018+COVID_3$Measles_2018+COVID_3$Expenditure_2016)
summary(Mod1)

Call:
lm(formula = log(COVID_3$Cases_million) ~ COVID_3$Year_2018 + 
    COVID_3$Measles_2018 + COVID_3$Expenditure_2016)

Residuals:
    Min      1Q  Median      3Q     Max 
-3.2724 -0.7267 -0.0046  0.8119  4.1725 

Coefficients:
                          Estimate Std. Error t value Pr(>|t|)    
(Intercept)              -6.779882   0.663669 -10.216  < 2e-16 ***
COVID_3$Year_2018        13.294002   0.837850  15.867  < 2e-16 ***
COVID_3$Measles_2018      0.011153   0.009456   1.179  0.24005    
COVID_3$Expenditure_2016  0.117851   0.042104   2.799  0.00579 ** 
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 1.273 on 152 degrees of freedom
Multiple R-squared:  0.7306,    Adjusted R-squared:  0.7253 
F-statistic: 137.4 on 3 and 152 DF,  p-value: < 2.2e-16

hist(Mod1$residuals)

shapiro.test(Mod1$residuals)

    Shapiro-Wilk normality test

data:  Mod1$residuals
W = 0.98965, p-value = 0.3087